Abstract

In this research study using Regression models of mixed Poisson for longitudinal data for analysis of sheep blood, longitudinal data was Known as evidence sectional measured in certain periods of time. These data gained in the current decade great importance, especially in the economic studies and medical.
This research includes mixed responses (discrete response represented by the distribution of Poisson and continuous response represented by the distribution of Gaussian) and these responses are correlated, thus when counting estimations of the parameter for models for every response separately, will give biased estimates, For this reason the joint regression models can be used for mixed responses by using the multivariate methods to estimate the parameter of the model, and these methods: method Factorization Model (FM) and Generalized Estimating Equation (GEE) has been using the simulation through a comparison between the two methods by statistical program (R) in addition to the application of two methods on real data for analizing of sheep blood.
The results of simulation obtained that they decrease at the MSE whenever the sample size increases, as that a paired in all methods that used in the study, and this reflects one of good properties when the value of estimator is approach from the actual value for the parameter at increase of the sample size, and had been noted that a big converge between the (GEE) and (Factorization) methods in the value of mean squared error.
From the results of simulation was applied at actual data to analize sheep blood that taken at intervals from 50 sheep's select a sample of blood from each sheep at 5 intervals and collected 250 subjects to study the effect of Enzyme (GPT) and (TRI) by some studied properties of animal sex, weight and type of blood (HB) for every animal and the results obtained that good estimations for the parameters of mixed models and there is correlation between the mixed response and independent variables. And they are accounted highly conduce between (FM) and (GEE) methods.